Using FISSA with Suite2p

suite2p is a blind source separation toolbox for cell detection and signal extraction.

Here we illustrate how to use suite2p to detect cell locations, and then use FISSA to remove neuropil signals from the ROI signals.

The suite2p parts of this tutorial are based on their Jupyter notebook example.

Note that the below results are not representative of either suite2p or FISSA performance, as we are using a very small example dataset.

Reference: Pachitariu, M., Stringer, C., Dipoppa, M., Schröder, S., Rossi, L. F., Dalgleish, H., Carandini, M. & Harris, K. D. (2017). Suite2p: beyond 10,000 neurons with standard two-photon microscopy. bioRxiv: 061507; doi: 10.1101/061507.

Imports

! Important note: if you want to run Suite2p and FISSA in the same python instance (as we do in this notebook), you have to make sure multiprocessing pools are started using the 'spawn' context, by having the following on top of your code: See here for more information: https://docs.python.org/3/library/multiprocessing.html

Run suite2p

Load the relevant data from the analysis

Run FISSA with the defined ROIs and data

Plot the resulting ROI signals

Note that with the above settings for suite2p it seems to have detected more small local axon signals, instead of cells. This can possibly be improved with manual curation and suite2p setting changes, but as noted above these results should not be seen as indicative for either suite2p or FISSA due to the small dataset size.

Also note that the above Suite2P traces are done without suite2p's own neuropil removal algorithm.